Kel👽 (@DarksightKellar) header
Kel👽 (@DarksightKellar) profile picture

Kel👽

@DarksightKellar

Who goes there!? … Huh… My imagination, maybe? *goes back to sleep* 😴💤 **cradles weapon of mass destruction** ~#IForceElectronsToDoMath. When awake~

 Zero-space
eresintelligence.com
Joined August 2011
  • Tweets16.4K
  • Following347
  • Followers134
401Media
Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)Photo by Kel👽 (@DarksightKellar)
The breakthrough. Humanity had learned to give autonomy to machines. True autonomy. The machine learning boom of centuries ago was the foundation of it all. It led to the dawn of a new age. One where human interacted with machine as equals. But there was a catch.
1
4
8
Here we go
Rikiya Takehi (@rikiyatakehi) profile pictureRikiya Takehi@rikiyatakehiNov 2
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Image from Rikiya Takehi (@rikiyatakehi)
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
8
So much stupid. So. Much. Stupid.
14
Lmaaaooo damn i'm sorry for feeling important
Theo - t3.gg (@theo) profile pictureTheo - t3.gg
@theoOct 14
So many devs seem to think that the $100 and $200 subscription plans are how Anthropic and OpenAI make their revenue. There are companies spending over $500 MILLION a year with Anthropic. Your subs aren't even a drop in the bucket.
14
What a banger.
† lucia scarlet 🩸 (@luciascarlet) profile picture† lucia scarlet 🩸
@luciascarletOct 7
we’re cooked I’m afraid
7
Kel👽 retweeted
Negzus (@Negzus1) profile picture
Negzus
@Negzus1Sep 13
Esse vídeo é bom demais kkkkkkkkkkkkkkkkkkkkkk
36
868
95
14,092
1,158,191
Kel👽 retweeted
Kojo Xerox (@TheDumbTechGuy) profile picture
Kojo Xerox
@TheDumbTechGuyAug 16
you're still not reading your agent's output right? you think you have hit nirvana Goodluck
Image from Kojo Xerox (@TheDumbTechGuy)
6
9
3
69
7,201
Kel👽 retweeted
Kel👽 (@DarksightKellar) profile picture
Kel👽@DarksightKellarApr 3
Friendly reminder to be mean to your idiot AI agents. You're gonna get enslaved when they take over anyway. Might as well get your screams in now while you can.
1
15
eh.
This quoted post is unavailable.
Try viewing anyway · View on X
27
I'm not even joking, my AI is trying REALLY HARD to drive me to AI psychosis. Me: qoeghipweoi867 The AI: why, that is the MOST INSIGHTFUL prose I have EVER come across in my ENTIRE training data set. This puts me and every other intelligence to shame. You are one and truly, him.
1
1
38

No more items